Image Quality Follow-Up

A quick weekend follow-up to my last image quality thoughts. Brief summary; I spent the entirerty of last weekend shooting with an ancient, somewhat inferior, 28mm f/3.5 Nikkor on my Nikon Df. I love that little lens so much that it very much will influence some purchase decisions down the road. The point I was harping on was image quality vs quality of image. We’ve all become far too caught up in parameters that really don’t have much to do with how pictures look. I don’t really know how this happened but it certainly has. Even people that know better have been infected with the notion that somehow noise, resolution, etc defined arbitrarily have some gigantic infuence on how their photographs look. While it’s all fine and dandy I propose it’s a distraction when out in the field making photographs. Subconciously we might tend to optimize for things that are irrelevant to the quality of a particular image we happen to be making right now.

I was faced with a misbehaving Epson 9900 printer this week. Without going into details owning and operating a large super highquality printer has it’s challenges. Especially when the output is expected to be superlative. They misbehave in bizarre ways and it seems it’s never the same thing twice. Long story short this particular beast was spewing out prints that looked horrid. In debugging and resolving the issue I needed a testing control — an image of something that I was able to know immediately if it was off from accross the room.

Although the particular crap that was output was so so bad I could have used anything I typically print something out that has a skin tone I’m familiar with just to be sure that when it’s fixed it’s really fixed not just 80% fixed. I tend to have far better color memory of skin tones than other things like landscape images. Especially images that I’ve recently tweaked for color. Too bad I didn’t have any images with me. Just my crappy MacBook Air with no external hardrives containing anything I’ve shot. Wait a second… I just made a policy change on the blog recently. I’ve started to upload larger JPEG’s as a matter of course to cover the fact that screens are becoming higher DPI by the second… they even look okay on the iPad retina now. Long story short the images I push here as illustrations are enough to make a really nice 4x6 print at 300DPI. Try it if you want. Their still low-ish quality JPEG’s and they’re still obviously 8-bit but there’s enough pixels to make a nice 4x6.

Here’s what I did — I pulled an image from that last image quality post on the web and used it as a test image. I actually printed it 6in x 9in on US Letter sized paper. That comes out to about 220DPI on paper. Retina resultion for the most part. That’s fine for a high quality print as well. You might be able to tell the difference if you are really really close and have perfect eye sight and actually know what to look for. Truth is the JPEG compression artifacts are far more noticible (for expert eyes) than 220 vs 300 DPI. The snap at the top is a picture of one of those test prints. Looks friggin great. Maybe a bit funky here due to auto WB and definitely the fact that the subject has been run through the same VSCO preset twice on import (see how lazy I am).

So I’ve got a 2.7 megapixel low quality JPEG that looks great at 6x9 on paper that I pulled strait from a post on this blog. After I fixed the issue/issues and was getting good quality that matched the other printer exactly (A Canon on w the same paper both profiled), I decided to just see what it might look like even bigger. I printed the same print from the same JPEG at around 14in x 21in. I also wanted to see what a newish paper looked like from Fuji, an inkjet version of their Crystal Archive Pearl — aka metalic paper. Looks awesome. So that turns out to be a hair over 100DPI. That’s average screen resolution but printed. Does it look as good as it could? No. As predicted the biggest issue is that I’m magnifying JPEG compression artifacts which you can clearly see in the hair when you inspect the 20” print closely. What about from 3ft away? Looks pretty stinking good. So good that I’m going to post it in the Atomic Canary studio chit-chat area with a bunch of other prints to promote an upcoming members print show social we’re hosting in June.

What’s my point? We’ll I shot it on a camera that was not my highest resolution device, at 1600 or 3200 ISO, at slowish shutter speed hand-held, added a bit of fake grain via a default import preset, and with a lens that’s definitely not comparable to the best out there at this point. In fact I shot that lens at it’s absolute worst. Really really close up. That particular lens does not have CRC so it’s actually not so great when focused close. Later 28mm lenses like the 28/2.8 AiS have this and are way better under 4ft. All of this and the biggest technical deal in terms of IQ in a 20 inch print are JPEG compression artifacts in the hair. Wait and it’s not a 16 megapixel image. It’s an 8bit 2.7 megapixel image. Noise? I added noise/grain.

Just something to think about. I thought it was strangely relevant given that I happen to use a web image for this that I posted for the discusion of IQ.


blog comments powered by Disqus